Pedigree Consultants : Thoroughbred Pedigree Analysts and Consultants

AN OPPORTUNE TIME

Alan Porter

It’s little more than a year since TrueNicks was launched. With its impact continuing to widen, and with the addition of the TrueNicks Broodmare Analysis Report to the stable of products, we’ve been fielding some inquiries from new users about the calculation of the rating and how this rating is turned into a score. So, we’ve taken the opportunity to revisit an explanation of the TrueNicks calculation, and particularly how we have tackled the potential pitfalls that can befall an opportunity-based system such as TrueNicks.

As I wrote when TrueNicks was launched, I’d first pondered the need for an opportunity basis for a valid assessment of Nicks and Sire Line Affinities more than 30 years ago (to be precise, in the June, 1976 issue of Stud & Stable magazine). After leaving full time journalism for full-time pedigree research and consultancy in mid-1980’s, I had always had the creation of some kind of opportunity based program in the back of my mind, and since coming to the U.S., I’d explored a number of options to put the plan into action, but had always been stymied by lack of access to a truly comprehensive database.

That changed around five years ago when The Jockey Club Information Services (TJCIS) came to an agreement in principal with Byron Rogers and me (collectively Pedigree Consultants LLC) to provide the data for a new nicking program. The TJCIS database is by far the most complete database of known foals, runners and race results from around the world and thus ideally suited our purpose of creating a nick rating based on the reality of what has actually been tried and is actually successful. We then took the concept and business plan to The Blood Horse Publications, who came on board as a partner in the development and distribution of the program. Thus was TrueNicks born.

As one might imagine, I’d had plenty of opportunity over the previous three decades to consider the methodology and problems in developing a true opportunity based system. Intuitively, it seems that one could obtain valid results by comparing the percentage of stakes winners bred on a cross to the percentage of horses bred on the cross, relative to the percentage of stakes winners in the population. This might form a useful comparison when a first generation cross has been tried a sufficient number of times, for example the current very popular cross of Unbridled’s Song over mares by Storm Cat, which has already produced seven stakes winner from 40 foals (superior to the performance of Unbridled’s Song, and of Storm Cat mares with all other mates, and superior to the population as a whole). More often that not though, we are dealing with instances where there are insufficient starters by a given sire, out of a mare by a given broodmare sire for valid results to be returned. It’s here that problems can begin for an opportunity based system, as was discovered in a study some years ago of the what was either the Bold Ruler/Princequillo or Nasrullah/Princequillo nick (whichever, the principle remains the same).

When we move from a first cross nick to second and third generation nicks, we begin to discover that some opportunities are more equal than others. When a sire, such as Bold Ruler, enjoys outstanding success (particularly as a sire of sires), sons with a wide variety of credentials are pressed into service as stallions. Those sons will tend bred to mares which emulate the most successful nick for their sire, and again this will involve mares with a wide variety of credentials. For example, moving from the Bold Ruler/Nasrullah era to a more recent cross, we might take Storm Cat/Mr. Prospector, which as a first generation cross produced 17 stakes winners from 73 starters (15% stakes winners to foals, a figure superior for Storm Cat with all other mares; Mr. Prospector mares with all other stallions; and the population as a whole). Since Storm Cat has been the most sought after sire of sires of his generation, tremendous numbers of his sons have retired to stud, many of whom have poor or no racetrack performance, and/or modest distaff pedigrees. Given the success of Storm Cat with Mr. Prospector mares, those less-credentialed sons of Storm Cat have covered a considerable number of Mr. Prospector mares with equally variable qualifications. As a result, while in pure statistical terms opportunity for the cross has considerably increased, the class level at which many of those crosses have taken place means that true opportunity (as defined potential for success in black-type competition) has not increased at the same rate. Therefore, it’s inevitable that in such a situation, the basic calculation of percentage of stakes winners bred on the cross, compared to percentage of starters bred on the cross rapidly trends back to, or even below, population norms (in both a commercial and population based sense). Of course, we were very aware of this potential pitfall when we created the formula for the TrueNicks rating. This is why our rating is calculated on completely different formula.

The calculation of the TrueNicks variant is a three step process. First, we take the percentage of stakes winners bred on the cross, and then compare that to the percentage of stakes winners produced when the specific stallion(s) responsible for the starters on the cross are bred to all other mares. For those that are comfortable with a Comparable Index (C.I) as opposed to an AEI, you will understand the methodology here. This gives us the SII (Sireline Improvement Index) figure. Second, we then take percentage of stakes winners bred on the cross, and compare that to the percentage of stakes winners produced when the specific mares that have produced runners on the cross are bred to all other stallions which gives us the BSII (Broodmare Sire Improvement Index) figure. Finally. it is then the multiplication of the SII and BSII that yields us the TrueNicks variant. It is important that it is understood that we are only looking at the groups of stallion(s) and mare(s) that are bred to one another, and how these two groups have done when mated to all other stallion(s) or mare(s) as the case may be. We are looking at how the cross has performed on a stakes winners to runners basis, thus considering the opportunity, and further qualify this performance by looking at how each group has done when mated with other sources ensuring a most accurate and relevant figure.

Since groups of sires and mares tend to be bred to mates of similar class (for example, mares that visit A.P. Indy are more likely to have been bred to Storm Cat, Distorted Humor, Kingmambo), and so on, and there is a strong tendency for distribution of crosses, and discrepancies in quality to be spread across the board (certain reasonable assumptions as to how populations behave have to be made for any nicking program to function). In this context there is actually the potential for a strong “nick” or affinity to have a greater impact at a lower level, giving it a shot at upgrading lesser quality breeding stock, than at a higher level, where sheer class is often as strong a factor as pedigree affinity. In addition, with TrueNicks looking at all starters bred on a cross, patterns tend to emerge within far fewer generations than if using smaller samples, thus dramatically reducing the proportion of irrelevant stock involved (more often than not a rating is produced without have to go back beyond the sire of the sire and the sire of the broodmare sire). When a rating has to be examined on the basis of a deeper cross, there is a tendency for it to involve relatively obscure or lower-achieving lines, and in those cases a stakes winner or two tends to have a much greater impact than in cases where a much larger population is involved.

Anyway, enough of theory, as all of this would be pretty academic if it didn’t work in practice. Before TrueNicks was launched, the precepts were tested, first in a feasibility study by TJCIS, and then in a comprehensive study of around 100,000 horses. The study of two complete foal crops yielded conclusive results, confirming that there was a strong statistical correlation between a high TrueNick rating and success at stakes level. In contrast to what might have been expected, this appears to run fairly consistently, without much variation across geographic region, generational distance, or class of horse.

When it came to representing the variant number (the raw number produced by multiplying the SII and BMSII), the distribution curve produced by the population provided some natural breakout points. A TrueNicks variant of 1.0 (indicating a stakes winner production rate exactly equal to opportunity, and thus average performance) fell very near the center of the population, as one might have hoped. Taking a slice of the population slightly above and below 1.0 gives us our ‘C’ or ‘average” rating. While only 13% of the entire Thoroughbred population were rated 2.5 or higher, nearly three times that percentage (37%) of stakes winners achieved that rating, so giving us a group that we felt comfortable to rate ‘A’ and upwards. Horses with a variant of 1.5 of and up represent just 30% of the entire population, yet provide more than three-quarters of stakes winners (77% to be precise). Thus the group rated 1.5 to 2.49 formed a natural group to rate ‘B.’

It’s intended that these proportions should remain relatively constant. Given that the proportion of stakes winners in the population remains relatively constant, it makes little sense to have program based on stakes production yielding an ever increasing proportion of highly-rated horses. To give an analogy: if one were trying to select a group of potential men’s Olympic 1500m. finalists by their best mile time, one might be looking for a standard that is surpassed by perhaps 2% or 3% of professional male middle-distance runners in any given year. A fixed standard, such as a four minute mile, might have done the trick in 1964, but would be of little use in 2009, when the once magic barrier has been breached by 40-year-olds and high schoolers. To be a useful selection tool, we would again have to find a standard that took us back to the 2% or 3% of the population we are looking for. Similarly, for a rating system to be a useful yardstick in decision making, the percentage of the horses rated at various levels needs to be kept both constant. This particularly applies to the higher levels. After all, if for example we designated 40% of the entire population were to be rated ‘A’ we wouldn’t done a great job of separating the wheat from the chaff and the value of a commerical application of the rating would be diminished.

We have, at various times made comment on three factors which bear repeating here. Firstly, we believe that the breakpoint where horses ratings should be taken as a clearly positive indicator is when a horse rates a “B+” or better on the scale. This breakpoint has been developed after looking at literally thousands of ratings on horses and is the point at where we feel that the odds truly swing in your favor. Secondly, we have maintained that TrueNicks requires some “intelligent interpretation.” While there is a strong statistical correlation between a high TrueNicks rating, and success at stakes level, the system is designed to be a guide to relative, rather than absolute, potential class. Indeed in the process of writing this post it has spurred some ideas for improvements in the system and we may eventually introduce some class measurements to create an overall predictive figure. We recognize that it is very possible that a “C” rated horse by a quality sire like A.P. Indy, will prove superior to an “A” rated horse by a $5,000 stallion. However, the “A” rated horse does represent an optimal mating for that particular sire, and all things being equal has a greater chance of success than a lower rated horse by the same sire, or for that matter, other stallions at the same commercial level.

Finally, TrueNicks was never created, nor intends to be “the answer”. We have maintained all along that the age old maxim that “paper doesn’t run very fast” applies here. While we believe that TrueNicks is an outstanding discriminator of potential stakes performance, the best in the marketplace today, it should not be used in absolutes, rather as a starting point or adjunct to other catalogue research, horsemanship and market savvy in order to get you into the winners circle.